Field of the Invention
[0001] This invention pertains to the field of semiconductor lasers.
Background of the Invention
[0002] As is well known, imperfections or defects (e.g., dislocations, point defects associated
with local areas of non-stoichiometry or foreign inclusions) that are present in semiconductor
lasers typically have the greatest effect on laser characteristics if the laser is
operated at high power level. Among such possible effects are such highly deleterious
ones as rapid deterioration of the laser, output power saturation caused by leakage
paths, and inhomogeneity of the distribution of the output envelope of the radiation.
Thus it is imperative that high power semiconductor lasers, including quantum well
lasers, be free of imperfections and defects to the greatest degree possible.
[0003] Quantum well lasers are considered to have many advantages over conventional double
heterostructure lasers. These advantages include improved high frequency response,
narrower linewidths, higher output power, reduced chirp under modulation, lower threshold
current densities, and increased temperature coefficient T
o. These advantages have been demonstrated in the GaAs/AlGaAs system.
[0004] An advantageous GaAs-based quantum well laser that comprises a graded index separate
confinement heterostructure has been reported. See D. Feketa et al.,
Applied
Physics
Letters, Vol. 49(24), pages 1659-60. This reference discloses GaAs-based strained layer quantum
well lasers, grown by MOCVD, wherein the 4 nm thick Ga
0.63In
0.37As quantum well is situated between two 0.2 µm thick Al
xGa
1-xAs regions having continuously graded refractive index, wherein x varies continuously
and linearly from 0 to 40%. The devices apparently have powers up to about 30 mW/facet,
and emitted at a wavelength of about 1 µm.
[0005] As those skilled in the art know, the AlGaAs system has the property that essentially
any composition within the system is lattice matched with any other composition within
the system. Due to this special property of the AlGaAs system, it is relatively easy
to produce a continuously graded index region of the type present in the Feketa et
al. lasers.
[0006] R. M. Ash, et al.,
Electronics Letters, Vol.25(22)'pp. 1530-1531(1988) disclose In P-based graded index separate confinement
heterostructure quantum well (GRIN-SCH QW) lasers with continuously graded confinement
layers of composition Al
yGa
0.48-yIn
0.52As, with y varying linearly from 0.35 to 0.25. The GaA1InAs system resembles the AlGaAs
system in that in the former the concentration in the growth atmosphere of only the
group III species need to be varied. However, other quaternary semiconductor systems
do not have this property. The InGaAsP is exemplary of these systems.
[0007] On the other hand, the InGaAsP system has pronounced advantages over the InGaAlAs
system that would make it highly desirable to be able to use the former in GRIN-SCH
QW lasers. In particular, the presence of Al, a strong oxygen getter, makes such highly
sensitive and complex devices as GRIN-SCH QW lasers relatively difficult to manufacture,
since it requires the vigorous elimination of all sources of oxygen in the growth
chamber. Under manufacturing conditions this is a difficult task, as those skilled
in the art well know, making GRIN-SCH QW lasers that do not comprise Al-containing
semiconductor material preferable to such lasers that comprise such material.
[0008] Due at least in part to the above referred to complication in the InGaAsP system,
it is generally believed by those skilled in the art that continuously graded index
regions are not feasible in this system (and in other multiconstituent systems wherein
the lattice constant depends on the composition).
[0009] In order to obtain emission wavelengths above about 1.2 µm in InGaAs-based QW lasers,
the quantum wells would have to be very thin (e.g., about 2-3 nm). Such thin wells
are difficult to make. Furthermore, laser performance might be hampered by the resulting
difficulties in the carrier capture process. On the other hand, these relatively long
wavelengths are of special interest for optical fiber communication. Of particular
recent interest are wavelengths at or near about 1.3 µm or 1.5 µm. For instance, radiation
at or near 1.48 µm can serve as pump radiation in a 1.55 µm optical fiber communication
system that comprises an Er-doped fiber optical amplifier. To be suitable as a source
of such pump radiation a laser should have relatively high output power, since the
gain that can be obtained in a given fiber optical amplifier increases with increasing
pump power. Exemplarily, such a laser should be capable of providing output power
of at least 10 mW/facet, preferably more than 25 or even 50 mW/facet. A. Kasukawa
et al.,
Japanese Journal of Applied Physics, Vol. 28(4), pp. L661-L663,
Electronics Letters, Vol. 25(2), pp. 104-105, and
Electronics Letters, Vol. 25(10), pp. 659-661 disclose 1.3 µm and 1.5 µm InP-based quantum well lasers
that comprise two step-wise graded GaInAsP confinement layers. Although the lasers
were reported to have been operated at quite high powers, the prior art devices contain
features that potentially can result in defects, and thus could reduce yield and/or
lead to decreased lifetime. In particular, the presence of step-wise composition changes
in the confinement layers can lead to these and other deleterious results. This is
due at least in part to the fact that stepwise compositional change requires the growth
of separate layers, and during the pause between layers (when source compositions
are changed) it is difficult to prevent defect formation at the interface.
[0010] In view of the potential importance of high power, long wavelength quantum well lasers
for, e.g., optical amplification in appropriately doped fibers, it would be highly
desirable to have available a quantum well laser that emits in the wavelength range
above about 1.2 µm (e.g., at a wavelength that is suitable for pumping of a fiber
optical amplifier), that can be readily manufactured and that can be operated at relatively
high power levels. As was discussed above, high power operation requires that the
laser be relatively free of imperfections and defects. Such a desirable laser therefore
would comprise design features that tend to reduce the incidence of imperfections
and defects. This application discloses such a laser.
Definition and Glossary of Terms
[0011] By "CW emission" we mean herein emission that has relatively constant amplitude over
periods of time that are sufficient to permit substantial establishment of thermal
equilibrium of the laser.
[0012] The "critical thickness x
c" herein is that thickness of a given strained semiconductor layer above which strain-relieving
dislocations generally appear. The critical thickness depends, inter alia, on the
composition of the given layer as well as on the compositions of the adjoining layer
or layers, and can be determined theoretically or experimentally. See, for instance,
H. Temkin, et al.,
Applied Physics Letters, Vol. 55(16), pp. 1668-1670.
Summary of the Invention
[0013] The invention is as defined by the claims. In a broad aspect the invention is apparatus
that comprises a semiconductor quantum well laser having an emission wavelength in
the range 1.2-1.68µm and being capable of room temperature CW emission, wherein the
laser comprises a confinement layer (typically two confinement layers, with one or
more quantum wells between them), that comprises GaInAsP, with the composition of
at least a significant portion of the material of the confinement layer varying continuously
through the thickness of the portion, with the compositional variation adjusted such
that the portion is substantially lattice matched. By "substantially lattice matched
"we mean herein that Δa/a ≦3x10⁻⁴, where
a is the lattice constant of the material to which the confinement layer is lattice
matched, and Δa is the maximum deviation of the lattice constant in the "lattice matched"
portion of the confinement layer from
a. Typically, the composition of both confinement layers varies continuously, with the
material of the confinement layer being single crystal InGaAsP.
[0014] Exemplarily, the confinement layer comprises a multiplicity of sublayers, with the
composition of a given sublayer varying essentially linearly with distance from an
appropriate interface. In currently preferred embodiments, the thickness is selected
to be less than the critical thickness for defect formation. This however is an optional
added precaution that may not be required in many cases.
[0015] Whereas the confinement layers in devices according to the invention are substantially
lattice matched and thus are not intentionally strained, the quantum well (or wells)
and, where applicable, the barrier layers between the wells, in these devices may
be, but need not be, strained. In fact, in some currently preferred embodiments the
quantum well composition is such that the wells are strained, with the well thickness
and composition selected such that the well thickness is below the critical thickness
for the given lattice mismatch. Desirably, the composition of the barrier layers between
the quantum wells is selected such that the strain in the former is the opposite of
that in the latter (e.g., compressive vs. tensile), such that the overall strain in
the region between the confinement layers is kept to a low value.
[0016] As is known, the band gap and optical properties of a quantum well can be varied
by means of lattice strain. Thus, by appropriate choice of lattice mismatch the range
of the well thickness-laser wavelength dependence can be extended to permit easy tailoring
of the laser wavelength through choice of growth parameters.
[0017] As discussed above, a significant aspect of lasers according to the invention in
the presence therein of a continuously graded, substantially lattice watched, InGaAsP
confinement layer. Even though this feature causes inventive lasers to be well adapted
for high power operation, inventive lasers need not necessarily be high power (i.e.,
more than 10, 25, or even 50 mW output power) lasers.
[0018] Apparatus according to the invention exemplarily is an optical fiber communication
system that comprises a length of rare earth (e.g., Er)-doped optical fiber, a laser
according to the invention that comprises means for causing an electrical current
to flow through the laser such that the laser has output radiation in the wavelength
range 1.2-1.68 (e.g., 1.48) µm, and means for coupling at least a part of the output
radiation into the rare earth-doped fiber. The apparatus of course also comprises
such conventional members as a source of signal radiation, means for coupling the
signal radiation into an optical fiber link that comprises the length of rare earth-doped
optical fiber, and means for detecting the signal radiation that was transmitted through
the fiber link.
[0019] Described is also a method of making a laser of the type described above. The method
comprises providing a III-V semiconductor body having a major surface, and exposing
the major surface to an atmosphere that comprises a multiplicity of molecular species
(comprising a first and a second species) such that a confinement layer is formed.
The step of forming the confinement layer comprises changing, during at least a part
of the formation time of the confinement layer, the concentration of the first species
and second species. The respective rates of change are selected such that the composition
of the resulting multiconstituent semiconductor material varies continuously with
distance from an appropriate interface, e.g., the major surface, with the lattice
constant of the resulting material being substantially constant.
Brief Description of the Drawings
[0020]
FIG. 1 schematically shows the layer sequence in an exemplary laser according to the
invention;
FIG. 2 shows exemplary data on gas precursor flow conditions for growth of InGaAsP
that is lattice watched to InP;
FIG. 3 gives data on the Ga content of an exemplary continuously graded InGaAsP layer;
FIGS. 4 and 5 show exemplary data on, respectively, the In content of InGaAsP and
the growth rate of InGaAsP; as a function of TMIn flow rate;
FIG. 6 gives an exemplary theoretical curve on the relationship between quantum well
composition and width required to achieve a particular laser wavelength; and
FIG. 7 schematically depicts exemplary apparatus according to the invention.
Detailed Description of Some Preferred Embodiments
[0021] FIG. 1 schematically shows the layer sequence in an exemplary preferred embodiment
of the invention, namely, an InP-based strained layer quantum well laser. Substrate
10 is S-doped (n-10¹⁸/cm³) 100-oriented InP, as is 1 µm thick epitaxial layer 11.
Lower confinement layer 12 consists of continuously graded InGaAsP and comprises 6
sublayers 121-126, each sublayer being about 40nm thick. Sublayer 121 at the interface
with 11 has a composition having a bandgap corresponding to λ
g = 0.99 µm (compositions in this quaternary system are frequently identified by a
corresponding bandgap wavelength. This notation is unambiguous and well understood
by those skilled in the art), with its composition being linearly graded to result
in composition corresponding to λ
g = 1.04 µm at the interface with sublayer 122. The compositions of all other sublayers
are similarly linearly graded, with the compositions at the other interfaces corresponding,
respectively, to 1.12 µm, 1.16 µm, 1.206 µm and 1.25 µm (at the interface with quantum
well 130). It is to be emphasized that at the interface between any two sublayers
there is no change in composition but only a change in the rate of change of composition.
In particular, there is a change in the rate of change of both Ga and As content with
distance from, e.g., layer 11. Strained quantum wells 130,131, and 132 each are 4.5nm
thick In
0.5Ga
0.5As, with 20nm thick InGaAsP layers 140 and 141 (of composition corresponding to λ
g = 1.25 µm) therebetween. Confinement layer 15 is the mirror image of 12, and is followed
by 50nm thick undoped (n-10¹⁶/cm³) InP layer 16,0.2 µm thick Zn-doped (p~4x10¹⁷/cm³)
InP layer 17, 0.9 µm thick Zn-doped (p~8x10¹⁷/cm³) InP layer 18,0.4µm thick Zn-doped
(p~1.1x10¹⁸/cm³) InP layer 19,0.12 µm Zn-doped (p~5x10¹⁸/cm³) InGaAsP (composition
corresponding to λg = 1.2 µm) layer 20, and by 0.06 µm thick Zn-doped (p~10¹⁹/cm³
InGaAsP (composition also corresponding to λ
g = 1.2 µm) layer 21.
[0022] The above-described epitaxial layer structure was grown by atmospheric pressure metal
organic vapor phase epitaxy (MOVPE). However, other growth techniques (e.g., low pressure
MOVPE, CBE or gas source MBE) can also be used to practice the invention.
[0023] The substrate 10 was prepared conventionally, and epitaxial layer 11 was grown thereon
by a conventional technique. The growth of the confinement layers was performed at
625°C, with trimethyl indium (TMIn) at 30°C, trimethyl gallium (TMGa) at -15°C, 5%
AsH₃ and 20% PH₃ as precursors. After dicing, provision of conventional optical coating,
electrical contacts, and heat dissipation means the laser output was measured. Output
in excess of 200mW was observed in a laser of structure substantially as described.
[0024] FIG. 2 shows exemplary data on the precursor flow conditions necessary to grow InGaAsP
layers that are lattice matched to InP, with λ
g ranging from 0.9 to 1.40µm. The data assumes that the flow of 70 sccm TMIn and 225
sccm of the 20% PH₃ is kept constant at a rate that results in excess In and P in
the growth atmosphere. The figure indicates, for instance, that, in order to obtain
material having a composition corresponding to λ
g = 1.01µm the TMGa and AsH₃ flow rates are to correspond to points A on curves 26
and 25, respectively. Analogous remarks apply to compositions corresponding to points
B, C, D, and E, as well as to all other points on the curves, as those skilled in
the art will recognize.
[0025] As discussed above, in devices according to the invention the composition of the
confinement layer is varied continuously over a considerable range, e.g., from λ
g = 0.99µm to λ
g = 1.25µm. This could be done by controlling the flow rate of TMGa and AsH₃ so as
always to lie on curves 26 and 25, respectively. This however would pose a complex
control problem (since curve 25 is highly non-linear) and is typically not necessary.
We have found that a piece-wise linear approximation gives perfectly acceptable results.
[0026] In order to grow a InGaAsP confinement layer of thickness t whose composition is
lattice matched to InP and varies continuously between λ
g1 and λ
g2, the quantity Δλ
g = λ
g2 - λ
g1 exemplarily is divided into n parts that define the compositions at which the rate
of change of TMGa flow and of AsH₃ flow will be changed. Exemplarily, these correspond
to points B, C, and D of FIG. 2.
[0027] By way of example, if the layer of material on which the InGaAsP confinement layer
is to grow has composition λ
g1 and if n = 6, then the following procedure can be used. By any appropriate means
(e.g., using FIG. 2) the TMGa and AsH₃ flow rates corresponding to λ
g1, λ
g2, and to the 5 intermediate compositions at which the rate of change of the flow rates
are to be altered are determined. Since the material growth rate at any given set
of flow rates (at a given constant temperature) is known, the time required to grow
any given sublayer can be readily determined. The sublayers will frequently be all
the same thickness. By determining the difference in TMGa flow rate corresponding
to λ
g1 and that corresponding to λ
g1 + Δλ
ga (the composition at which the rate of change of the flow rates is to be changed for
the first time), and dividing the thus determined difference in flow rate by the previously
determined time to grow the first sublayer, the appropriate rate of change of TMGa
flow rate is determined. By an analogous procedure the appropriate rate of change
of AsH₃ is determined. This procedure can be applied in analogous fashion to determine
the appropriate rates of change of flow rates for all the other sublayers of the desired
InGaAsP confinement layer.
[0028] Computer controllable mass flow controllers suitable for use with TMGa and AsH₃ are
commercially available, and software to program a computer to change the flow rates
in the above-described fashion is routine for those skilled in the art and does not
require discussion.
[0029] Continuing with the discussion of the exemplary embodiment; after the layer of composition
λ
g1 has reached the desired thickness, the flows of TMGa and AsH₃ are changed, without
growth interruption, such that the flows vary at the respective, previously determined
new rates. After elapse of the previously determined time, the rate of change of the
respective flows are again adjusted without interruption of the growth. This procedure
is continued until all the sublayers are completed.
[0030] Since the inventive procedure for growth of InGaAsP confinement layers requires only
changes in the rate of change of gas flow, the growth of a confinement layer can be
continuous, without cessation of growth upon completion of a given sublayer, as necessarily
occurs in the growth of prior art step-wise graded InGaAsP confinement layers. This
is considered to be a significant aspect of the invention, since defects tend to easily
form from the ambient in the absence of growth.
[0031] FIG. 3 shows exemplary data on the Ga content of a In
xGa
1-xAs
yP
1-y continuously graded confinement layer according to the invention, with y-2.2x in
the solid, and λ
g varying from 1.1µm at arrow 60 to 1.25µm at arrow 61. As can be seen, the Ga concentration
varies continuously (linearly over a major portion of the layer thickness), thereby
reducing the likelihood of defect formation.
[0032] Upon completion of the growth of the lower confinement layer (e.g. 12 of Fig. 1)
typically a quantum well is grown, preferably without (or with minimal) pause in the
growth. The quantum well advantageously is In
xGa
1-xAs, with the value of x and the well thickness chosen to yield the desired lasing
wavelength. FIG. 4 shows exemplary data on the In content of InGaAs as a function
of TMIn flow rate. As can be seen, the dependence is linear, facilitating straightforward
adjustment of the solid composition.
[0033] Fig. 5 shows exemplary data on the growth rate of InGaAs as a function of TMIn flow
rate (with TMGa and AsH₃ flow fixed at 9 and 160sccm, respectively). The dependence
is linear, with the growth rate under the exemplary conditions depending only on the
TMIn flow rate.
[0034] For at least some applications of lasers according to the invention (e.g., for use
as a pumping source of a Er-doped fiber amplifier) it is necessary to finely tune
the laser wavelength. This is made possible through adjustment of a degree of strain
in the quantum well or wells. FIG. 6 shows an exemplary theoretical curve on the relationship
between x (in In
xGa
1-xAs) and the well width needed in order to obtain a lasing wavelength of 1.48µm. Experimental
data we have obtained are in good agreement with the theoretical curve, at least for
x≧0.43. For x = 0.53 the composition is lattice matched with InP.
[0035] FIG. 7 schematically depicts exemplary apparatus according to the invention, namely,
optical fiber communication system 70. Signal laser 71 emits signal radiation 72 which
is coupled into standard transmission fiber 73 and transmitted therethrough to fiber
amplifier 76. Pump laser 74 emits pump radiation of an appropriate wavelength (e.g.,
1.48µm) which is transmitted by means of a short length of fiber 75 into 76. The latter
comprises a rare-earth doped fiber 81 coupled signal-transmissively, by means of conventional
connectors 79, to 73 and to transmission fiber 77. The amplifier 76 also comprises
means 80 to couple the pump radiation from 75 to 81. Such means can be conventional.
Signal radiation is amplified in 76 in known manner, and is then transmitted through
77 to detector 78.